Freeze-in, glaciation, and UV sensitivity from light mediators

TL;DR

This paper investigates how initial conditions of dark radiation affect dark matter freeze-in via light mediators, revealing significant dependencies and extending the traditional freeze-in paradigm.

Contribution

It introduces the concept of glaciation scenarios, derives an exact collision integral for different temperature species, and explores the impact of initial dark radiation conditions on DM yield.

Findings

DM yield depends significantly on initial dark radiation temperature.

The glaciation band can extend the traditional freeze-in region by an order of magnitude.

DM phase space distribution is strongly influenced by initial conditions.

Abstract

Dark matter (DM) freeze-in through a light mediator is an appealing model with excellent detection prospects at current and future experiments. Light mediator freeze-in is UV-insensitive insofar as most DM is produced at late times, and thus the DM abundance does not depend on the unknown early evolution of our universe. However the final DM yield retains a dependence on the initial DM population, which is usually assumed to be exactly zero. We point out that in models with light mediators, the final DM yield will also depend on the initial conditions assumed for the light mediator population. We describe a class of scenarios we call "glaciation" where DM freezing in from the SM encounters a pre-existing thermal bath of mediators, and study the dependence of the final DM yield on the initial temperature of this dark radiation bath. To compute DM scattering rates in this cosmology, we derive for the first time an exact integral expression for the Boltzmann collision term describing interactions between two species at different temperatures. We quantify the dependence of the DM yield on the initial dark temperature and find that it can be sizeable in regions near the traditional (zero initial abundance) freeze-in curve. We generalize the freeze-in curve to a glaciation band, which can extend as much as an order of magnitude below the traditional freeze-in direct detection target, and point out that the DM phase space distribution as well as the yield can be strongly dependent on initial conditions.